This application proposes to develop and implement an interactive voice response (IVR) program to obtain self-report and neuropsychological data from 120 treatment-seeking patients that is predictive of alcohol treatment drop-out with six weeks of treatment intake. The proposed research will compare the neuropsychological data collected by IVR to data obtained using computer software that has previously been validated. Machine learning algorithms will be applied to the IVR data to derive models predictive of treatment drop-out. The generalizability of these models will be evaluated using an independent validation sample. PROPOSED COMMERCIAL APPLICATIONS: Substance abuse treatment trends toward increased treatment durations with decreased intensity cannot succeed if patients discontinue treatment soon after initiation. An inexpensive, telephone driven computer system that provided valid prognostic information offers both the providers and payors of services opportunities to develop and enhance more effective treatment programs in an increasingly competitive marketplace.